LexisNexis Risk Solutions has introduced Emailage Adaptive, a self-calibrating artificial intelligence fraud scoring system designed to learn continuously from an organization’s transaction data and confirmed fraud outcomes.
The product arrives as financial institutions and digital businesses face faster-changing fraud patterns and increasing pressure to reduce friction for legitimate users. TNGlobal has previously covered LexisNexis Risk Solutions research on rising digital attack activity and the growing use of AI in fraud and identity systems.
Emailage Adaptive analyzes signals including email, IP address, phone and address data and builds tailored models using an organization’s own fraud feedback and industry context.
Models recalibrate without manual tuning
The company said the system automatically recalibrates as new fraud patterns are confirmed, reducing the lag associated with manually updating fraud models.
In testing cited by LexisNexis Risk Solutions, the system captured around 90 percent of fraud among the highest-risk transactions while reducing false positives by more than 80 percent. Those results come from company testing and should not be treated as universal performance guarantees across every deployment.
The platform also provides reason codes intended to show which factors influenced a risk score. LexisNexis Risk Solutions positions this explainability layer as a way for fraud teams to understand automated decisions rather than relying only on a black-box score.
Adaptive fraud models depend on feedback quality
Continuous recalibration can reduce manual model maintenance, but it also increases the importance of the data and labels used to train the system. If fraud outcomes are incomplete, delayed or incorrectly classified, an adaptive model may learn from weak feedback.
LexisNexis Risk Solutions said Emailage Adaptive relies on a regular flow of customer transaction data, industry trends and substantiated fraud feedback. The system is designed to use those inputs to refine its scoring over time.
That makes governance around data quality, review thresholds and model monitoring an important part of any deployment, especially where automated risk scores affect account opening, payment approval or other consequential decisions.
Fraud teams are trying to reduce both losses and friction
Fraud controls increasingly have two competing goals: catching more malicious activity while avoiding unnecessary declines or manual reviews for legitimate users.
Adaptive models are one response to that problem because they can be updated more frequently than traditional rule sets or manually tuned models. Their practical value, however, depends on whether improvements persist across changing transaction volumes, geographies and fraud types.
LexisNexis Risk Solutions launched the product globally in September. The company said the system is designed for organizations that want more automated risk scoring without repeatedly rebuilding fraud models by hand.